Detection of trace emerging contaminants in seawater by POCIS: optimization of the sampler processing to enhance the accuracy of the LC-MS/MS analysis

The detection of emerging contaminants (ECs) in marine water represents a challenge, due to the huge dilution factors, the variety of their physicochemical properties and the complexity of the matrix. The application of methods based on classical spot sampling may not be sufficient to detect the expected concentrations (trace and ultra-trace levels). A valid alternative is constituted by passive sampling, combined with sensitive analytical techniques such as liquid chromatography-tandem mass spectrometry (LC-MS/MS). The passive sampling approach allows to enhance sensitivity thanks to the in-situ pre-concentration, which permits to virtually sample rather high volumes of water, during a deployment period (days to weeks of exposure). Through relatively long deployments, integrative sampling is possible, thus permitting to obtain time-weighted average concentrations and identify potential episodic events, undetectable by spot sampling. A passive sampler particularly suitable to detect mid-polar analytes is the Polar Organic Chemical Integrative Sampler (POCIS), in which the hydrophilic lipophilic balance (HLB) sorbent is sandwiched between two polyethersulfone membranes [1]. Several protocols were proposed in the literature for the treatment of the POCIS after retrieval, generally involving the packing of the phase in a cartridge, a washing step and solvent elution [2]. However, most works deal with the deployment of POCIS in freshwater. Due to the substantial difference between fresh and sea waters, a fit-for-purpose processing of POCIS deployed in the marine environment is desirable, in order to obtain reliable data. In fact, the presence of salts and biofouling strongly affects the analytical signals, making it necessary to optimize the protocol to limit the matrix effects and maximize the recovery of the compounds. In our study, 22 emerging contaminants, including UV-filters, pharmaceuticals, tracers and perfluorinated compounds, were investigated. After optimizing the chromatographic separation and mass detection, several tests were performed to obtain the best processing strategy in terms of analysis accuracy. In particular, three different solvents (methanol, dichloromethane:isopropanol, 80:20 (v,v) and acetone) were tested as eluents on a model system, in which the analytes were spiked on the HLB sorbent, previously left in contact with synthetic seawater for 7 days. The results pointed out that the most polar compounds were lost during the washing step and strongly affected by ion suppression. Nevertheless, no significant differences were observed among the three eluents neither in terms of recovery nor matrix effects. A verification on three field POCIS, deployed in the same site (harbour waters) for 21 days, was performed. The three protocols showed a different performance on the exposed samplers: low recoveries of anionic compounds were observed when acetone was used, while UV-filters were completely eluted only by the dichloromethane:isopropanol solvent mix. The presence of interferent compounds, apart from salts, probably caused a different elution behaviour of the analytes in the field POCIS compared to the model system. From these results we proposed a final method consisting of a two steps elution on the packed sorbent, with methanol and dichloromethane:isopropanol, 80:20 (v,v). The overall method showed good procedural precision (relative standard deviation ranging from 1 to 28%), average recoveries of 54-140% (except the most polar analytes) and excellent limits of detection, namely 0.05-9 ng per g of HLB phase. Some POCIS deployed in the Ligurian coast during summer 2021 were processed with the optimized method, which allowed the detection of 16 out of the 22 target compounds. Caffeine, ketoprofen and the UV-filter octocrylene were detected at the highest concentrations (approximately 100-300 ng g-1 of HLB phase); the other analytes, namely paraxanthine, perfluorinated compounds, other pharmaceuticals and UV-filters, ranged from “non-detected” to 100 ng g-1 of HLB phase. By comparing POCIS deployed in duplicate, satisfactory reproducibility of the complete procedure (sampling plus processing) was observed: the relative standard deviation (RSD) for the detected analytes ranged from 8 to 30%, except for some more lipophilic substances, which were characterized by an RSD of approximately 50%. Further optimization will be needed in order to enlarge the list of target compounds when using POCIS in marine waters, to improve the reliability in the detection of the most lipophilic and hydrophilic compounds. However, the optimized protocol was demonstrated to be suitable for the accurate quantitation of several emerging contaminants in POCIS deployed in seawater.